Maelstrom wrote:Call it a project for 1000 foo. When you're done, you get anywhere between 800 to 1200 foo, and there's no distribution because it's a single occurence. Make that a project for 50 foo 20 times, and you're going to get very close to 1000 foo in the end, because each event has a random component. Some will be high, some low, but they should tend to balance out. If you could slice the project infinitely small and into an infinite number of repeats, you would get exactly 1000 foo.
I'm just wondering whether you're assuming that we implemented a normal distribution, or assuming that if you make random draws between 800 and 1200 you will automatically get a normal distribution
... The latter is false, but I suppose you assumed that we implemented it. For as far as I remember, we did, so your analysis would be right.
For different projects, that is - the deviation is calculated when the project is created, so repeating projects have a fixed deviation.
